Research on the application of Wireless video information mining for the multi-encoding mode and trans-coding optimization algorithm

The basic idea of the video data mining is based on the content of multimedia features and the semantics related to these properties, from large multimedia data sets and analysis of the underlying discovery, effective, valuable, and understandable patterns. In this paper, based on multi-encoding mode for real-time trans-coding of video streams we firstly research on the input data reuse pattern matching and mining selection algorithm (hidden Markov model), analyzing the input video stream within four blocks of the same DCT coefficients and variance of the weighted average. At the same time, we study the rate distortion optimization algorithm based on hierarchical model, in the frame layer and macro-block layer resynchronization marker insertion and intra-coded macro block updates of different coarse grained rate-distortion optimization. In the establishment of the frame layer rate-distortion model, the full account number on the rate of intra-coded model of the model and derive a concrete formula for the frame layer of the Lagrange multiplier. Finally, we corrected Lagrange multiplier focus on the characteristics of the codec for the macro block layer. Wireless video information mining and trans-coding optimization algorithm have important application values in the field of wireless video communication such as Wireless video surveillance, mobile TV, mobile video telephony conferencing, mobile e-commerce and so on.